Infinite Game OS

Definition

Grant Researcher takes an operator profile and a cycle window and produces a ranked list of grants whose mission, eligibility, deadline runway, ask-size, funder track record and capacity match the operator's actual work. The default source mix queries Grants.gov via the public search2 endpoint (federal opportunities, no auth), a configured web search engine (foundation, state, local, corporate, RFP feeds) and ProPublica Nonprofit Explorer (peer organizations' 990 filings as funder discovery). Optional paid databases (Foundation Directory Online, Instrumentl, GrantStation) layer in when the operator subscribes. Fit scoring runs a six-component rubric (mission alignment, eligibility as gating, deadline runway, ask-size, funder track record, capacity). Disqualifiers are gates not warnings: missing eligibility, mismatched ask-size, scope distortion that would bend the work to fit the funder, missing SAM.gov registration for federal applications. The skill's highest-leverage function is fast disqualification, helping the operator say no quickly to poor-fit opportunities so writing capacity goes to high-fit ones.

Profile as the spine, fit score as the ranking

The operator profile is the spine of every fit decision downstream. Required fields: mission, org type (501c3, LLC, individual, fiscally sponsored, public agency), primary geography, focus areas, prior grant history, ask-size range, current capacity, eligibility constraints (citizenship, certifications, fiscal sponsorship status, indirect cost rate). The reference format is markdown with frontmatter for structured fields and a prose body for nuance. Without the profile, fit scoring has no inputs; the skill halts at Step 1.

Fit scoring runs a six-component rubric: mission alignment (0-30), eligibility as gating (0-25), deadline runway (0-15), ask-size fit (0-10), funder track record (0-10), capacity match (0-10). Disqualifiers act as filters not penalties: a missing eligibility requirement, an ask-size well outside funder norms, a deadline runway too short for a competitive proposal, a federal opportunity without active SAM.gov registration. The skill's job is fast disqualification first, ranked recommendation second. The operator's writing capacity gets protected before the funnel expands.

Discovery versus Fit Deep-Dive

Discovery mode is the cold-start sweep. It queries the configured sources within a cycle window, scores each opportunity, surfaces a top-10 ranked list and below it a disqualified-with-reasons block so the operator can see what was filtered out. Best for new operators entering a funding landscape, established practitioners running quarterly landscape refreshes or a domain shift requiring a fresh source mix.

Fit Deep-Dive mode runs against one named opportunity. The output is a single memo covering eligibility analysis, mission alignment, ask-size and capacity, deadline and effort estimate, funder track record from public records, three risk factors and a recommended next move with reasoning. Best for an experienced practitioner who already has a known universe and wants a fast yes/no on a specific RFP, or a board ask for a fit memo before committing capacity.

Use Cases

Quarterly grant landscape refresh for an established nonprofit

A 501c3 with five years of grant history wants to surface adjacent funders before allocating Q3 writing capacity. Discovery mode sweeps Grants.gov, web search and the operator's configured paid databases against the 12-month cycle window. The top 10 ranked opportunities arrive with fit scores, why-it-fits bullets and suggested next moves. The disqualified-with-reasons block surfaces where filters might be over-aggressive.

Fast yes/no on a specific federal RFP

A program officer recommends a federal opportunity. The operator wants a fit decision before committing writing capacity. Fit Deep-Dive mode runs against the named opportunity, produces a single memo with eligibility analysis (including SAM.gov registration check), mission alignment, ask-size analysis, deadline runway and three risk factors. The memo lands in under an hour; the operator decides in minutes.

New nonprofit entering the funding landscape

A newly-incorporated 501c3 wants to see what is out there for their domain. Discovery mode sweeps wide. The disqualifier block teaches the operator which kinds of opportunities to filter against (eligibility mismatches, federal opportunities requiring SAM.gov registration the operator hasn't completed, ask-sizes well outside the operator's capacity). The first run is partly a calibration exercise; the second run, weeks later, is materially more efficient.

Funder relationship discovery via peer 990s

An experienced practitioner wants to find foundations that fund peer organizations. The skill queries ProPublica Nonprofit Explorer for the operator's named peers, extracts their funder lists from 990 filings and surfaces funders that appear across multiple peers. The output ranks by frequency-of-funding and ask-size match. Warm-relationship pipelines convert at materially higher rates than cold discovery; this is how the warm pipeline gets seeded.

Quarterly board memo on the funding landscape

A board asks "what is out there for our work?" before the next planning meeting. Discovery mode produces the ranked list. The operator pairs it with a brief board memo summarizing the top 3-5 opportunities, the disqualified opportunities worth surfacing as not-yet-fit-but-watching and any structural shifts in the funder landscape (new program launches, board changes at named funders, geography expansions).

FAQ

Why a skill instead of just searching Grants.gov directly?

Grants.gov gives raw opportunities; the skill gives ranked fit. The profile-load gate, the six-component scoring rubric and the disqualifier filter are what convert "100 opportunities returned" into "5 to 8 worth writing." Searching directly is the cheap part; the skill's contribution is the structural discipline that protects writing capacity.

Do I need a paid database subscription?

No. The default source mix runs on free public data: Grants.gov for federal, web search for foundation discovery, ProPublica Nonprofit Explorer for 990 lookups. Foundation Directory Online and Instrumentl layer in if the operator subscribes; many public libraries offer free FDO access. The skill works either way; depth scales with the source mix.

How does the sovereignty gate work?

The disqualifier list includes a gate question: would the operator do this work without this grant? If the answer is no, the skill surfaces the tension explicitly rather than scoring the opportunity. Grants compound when they fund work that was going to happen anyway. They become extractive when the program exists only because the grant exists.

What if I already know my universe of funders?

Run Fit Deep-Dive mode against specific opportunities rather than Discovery sweeps. The skill's structural value applies equally to known opportunities: eligibility analysis, ask-size fit, deadline runway, funder track record from 990 records. Experienced practitioners often run Fit Deep-Dive most of the time and Discovery quarterly to surface adjacent funders.

Can this skill submit applications?

No. Researcher discovers and analyzes; it never auto-submits. Submission lives in the paired Grant Manager skill, where the operator confirms the package before any submission action. The boundary protects the operator from auto-submitting a proposal whose final pass they didn't personally review.